Python机器学习实战：交叉验证

在 Python 中，可以使用scikit-learn库来实现多种类型的交叉验证。

1. K 折交叉验证（K-Fold Cross Validation）

K 折交叉验证是将数据集分成 K 个互不重叠的子集，每次将其中一个子集作为测试集，其余 K - 1 个子集作为训练集，重复 K 次，最后将 K 次的评估结果进行平均。

from sklearn.model_selection import KFold
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import make_classification
from sklearn.metrics import accuracy_score
import numpy as np

# 生成示例数据集
X, y = make_classification(n_samples=100, n_features=10, n_informative=5, n_redundant=0, random_state=42)

# 创建K折交叉验证对象，这里K设为5
kf = KFold(n_splits=5, shuffle=True, random_state=42)

# 创建逻辑回归模型
model = LogisticRegression()

# 存储每次交叉验证的准确率
accuracies = []

# 进行K折交叉验证
for train_index, test_index in kf.split(X):
    X_train, X_test = X[train_index], X[test_index]
    y_train, y_test = y[train_index], y[test_index]
    
    # 训练模型
    model.fit(X_train, y_train)
    
    # 预测
    y_pred = model.predict(X_test)
    
    # 计算准确率
    accuracy = accuracy_score(y_test, y_pred)
    accuracies.append(accuracy)

# 输出每次交叉验证的准确率和平均准确率
print("每次交叉验证的准确率:", accuracies)
print("平均准确率:", np.mean(accuracies))

2. 分层 K 折交叉验证（Stratified K-Fold Cross Validation）

当数据集中各类别的样本数量不均衡时，使用分层 K 折交叉验证可以保证每个子集内各类别的比例与原始数据集相同。

from sklearn.model_selection import StratifiedKFold
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import make_classification
from sklearn.metrics import accuracy_score
import numpy as np

# 生成示例数据集
X, y = make_classification(n_samples=100, n_features=10, n_informative=5, n_redundant=0, weights=[0.2, 0.8], random_state=42)

# 创建分层K折交叉验证对象，这里K设为5
skf = StratifiedKFold(n_splits=5, shuffle=True, random_state=42)

# 创建逻辑回归模型
model = LogisticRegression()

# 存储每次交叉验证的准确率
accuracies = []

# 进行分层K折交叉验证
for train_index, test_index in skf.split(X, y):
    X_train, X_test = X[train_index], X[test_index]
    y_train, y_test = y[train_index], y[test_index]
    
    # 训练模型
    model.fit(X_train, y_train)
    
    # 预测
    y_pred = model.predict(X_test)
    
    # 计算准确率
    accuracy = accuracy_score(y_test, y_pred)
    accuracies.append(accuracy)

# 输出每次交叉验证的准确率和平均准确率
print("每次交叉验证的准确率:", accuracies)
print("平均准确率:", np.mean(accuracies))

3. 留一交叉验证（Leave-One-Out Cross Validation, LOOCV）

留一交叉验证是 K 折交叉验证的一种特殊情况，其中 K 等于样本数量，即每次只留一个样本作为测试集，其余样本作为训练集。

from sklearn.model_selection import LeaveOneOut
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import make_classification
from sklearn.metrics import accuracy_score
import numpy as np

# 生成示例数据集
X, y = make_classification(n_samples=10, n_features=10, n_informative=5, n_redundant=0, random_state=42)

# 创建留一交叉验证对象
loo = LeaveOneOut()

# 创建逻辑回归模型
model = LogisticRegression()

# 存储每次交叉验证的准确率
accuracies = []

# 进行留一交叉验证
for train_index, test_index in loo.split(X):
    X_train, X_test = X[train_index], X[test_index]
    y_train, y_test = y[train_index], y[test_index]
    
    # 训练模型
    model.fit(X_train, y_train)
    
    # 预测
    y_pred = model.predict(X_test)
    
    # 计算准确率
    accuracy = accuracy_score(y_test, y_pred)
    accuracies.append(accuracy)

# 输出每次交叉验证的准确率和平均准确率
print("每次交叉验证的准确率:", accuracies)
print("平均准确率:", np.mean(accuracies))